It seems like Jeff Kaplan is always on the Overwatch forums. I don’t know how much work he’s getting done considering how often he posts and to the great extent that he does so.
This time around Kaplan has weighed in on matchmaking and MMR, and by “weigh in,” I mean seriously. If words were pounds, this post would weigh nearly 2,700 lbs.
It’s an interesting read, and an interesting reply to an interesting post. Seems as though a lot of cogs are at work in the background to match you the best way possible. I’d previously heard from a few random people while playing that your win/loss ratio went more into matching you than any sort of rating. That seemed odd, but seeing as up until today Blizzard hadn’t really talked much about their system, there was nothing wrong with baseless speculation, but baseless speculation it was, and it was completely wrong. There’s a lot of calculations and considerations that go into matchmaking in Overwatch, and your win/loss ratio isn’t one of them. In fact, it’s completely irrelevant. But there’s a lot of other factors that go into matchmaking.
The biggest one, though, is time. Overwatch’s matchmaker will try to find you the best match, but it also doesn’t want you to wait too long. This is especially important to players with a really high (or low) MMR, or if you’re queued with a six-stack. Those are especially hard to match since the players will be of vastly different pings and skill levels. Overwatch doesn’t want to prevent you from playing with your bad friend on the other side of the world if that’s what you want.
Much like time, another big factor is ping. It tries to match you with players as close as possible to your location to minimize those netcode issues that have become so pervasive. Despite how big of an issue the “netcode” in Overwatch seems to be, I think a lot of people still don’t realize these issues will always be issues, at least until we invent quantum-entangled internet or something. Even if Overwatch rose its tickrate to well over 100, there would still be inconsistencies, because information can only travel so fast (the speed of light).
Then you have the MMR itself. A lot of factors contribute to how much you gain or lose upon winning or losing. Blizzard has a butt-load of metrics on the maps, and their individual win/loss ratios, and even that gets taken into account. If you lose while defending on a map that favors offense, then you’ll lose less MMR than if you lose while attacking on the same map. This just further details how useless the win/loss ratio is in the game, and it’s probably something you really shouldn’t concern yourself with.
One thing that MMR can’t and doesn’t account for is your skill level with individual heroes. I know there are heroes that I’m really good with, and others that I’m completely trash. It’s unfortunately that the MMR doesn’t take this into account somehow, but I can understand why it doesn’t.
Also interestingly enough, players’ MMR carried over from the closed and open beta. That’s why sometimes you see low leveled characters matched with high leveled characters. Firstly, level doesn’t matter (remember that), and secondly, that low level character might not be new to the game, just new to release. Even if levels don’t matter, I would expect someone who is well over 100 to have more experience and therefore be potentially better at the game than fresh players. Overwatch also tries to segregate new players to give them their own environment. So that level 1 player who just happened to stumble into your game probably isn’t a bad player and probably shouldn’t be treated as such. For all you know, they’ve played 1000s of hours in the closed beta and were simply a little late to the party.
Kaplan also mentioned that the avoid feature has been straight-up disabled, and will be removed in a later patch. Evidently it was wreaking havoc with the matchmaking system. Really good players were being habitually avoided, and then they were having a hell of a time finding games. He still urges players to report/block toxic players, and they will be dealt with eventually, but the avoid feature was doing more harm than good.
I can safely say that I understand Overwatch’s matchmaking system much better than I did two hours ago, and that’s almost entirely thanks to Jeff Kaplan writing up a veritable dissertation. He continues to impress with his transparency, and it’s safe to say that Overwatch is in good hands.